GA4 Audit for Local Business: 12 Checks and a Prompt
November 17, 2026·Nataliia Rumiantsewa· 11 min read All posts
A GA4 audit is a short review of the settings that decide whether the numbers can be trusted. For a local business the review takes about an hour and covers 12 settings: data retention, key events, referrals, filters, time zone, currency and the Google Ads link. The most expensive mistake sits in the first one. A standard GA4 property keeps event data for 2 or 14 months, never longer.
This guide lists the 12 checks in the order they are best done, shows where each setting lives, and ends with a prompt that turns the list into a pass-or-fail report. Every limit comes from Google's help pages, listed under Sources. The worked examples use labelled assumptions, not client data.
14 months→
Longest event retention, standard property
Google Analytics Help
50→
Unwanted referrals per data stream
Google Analytics Help
400→
Google Ads links per property
Google Analytics Help
48 hours→
Wait for Ads data in GA reports after linking
Google Analytics Help
Source: Google Analytics Help, checked 17 November 2026
Why Audit GA4 at All?
Because GA4 never warns about a bad setting. Reports keep filling with numbers, and a wrong number looks the same as a right one. A short audit catches three problems that cost local businesses the most: history that disappears, conversions that are not conversions, and traffic credited to the wrong source.
14 months
The longest event-data retention available on a standard GA4 property. Options are 2 or 14 months; 26, 38 and 50 months exist only in Analytics 360.
Google Analytics Help: Data retention, checked 17 November 2026
Run the audit once after setup, again after any site or ad change, and then every quarter. Each check below names the screen to open, what a good setting looks like and what breaks when it is wrong.
How Long Does GA4 Keep Your Data?
Check 1: event data retention. Open Admin, then Data settings, then Data retention. A standard property can keep user-level and event-level data for 2 months or 14 months. The longer 26, 38 and 50 month options are limited to Analytics 360. Pick 14 months. Saving the change needs the Editor role.
The setting matters for explorations and funnel reports only. Standard aggregated reports are not affected. Age, gender and interest data keep just 2 months whatever the choice. A change applies after a 24-hour wait, and a switch to 14 months also covers data already collected and not yet deleted.
Check 2: reset on new activity. The same screen has a switch called Reset user data on new activity. When it is on, every new event from a user restarts that user's retention clock. When it is off, the identifier expires after the set period. Decide on purpose and write the choice down. The switch affects user-level data only.
Assumption example. A salon wants to compare November with last November in an exploration. The property still has the 2-month setting. The exploration can look back about 60 days, so the yearly comparison is impossible. Switching to 14 months fixes future comparisons only, because deleted data does not return.
Are Your Key Events Set Up Right?
Check 3: a real key event exists. A key event measures an action that matters to the business, such as a booking, a call tap or a form submit. Any collected event can become one. Open Admin, then Events, and look at the key event switch. If the list holds only automatic events, nothing in the reports measures money actions.
Check 4: the thank-you page, not page_view. The page_view event fires on every page, so marking it as a key event counts every pageview as a conversion. Google's documented route is to create a new event from page_view with the thank-you page URL, name it for example thank_you, and mark that new event. The new event can take time to show up in the Recent events tab.
Check 5: no event was overwritten. Google separates creating an event from modifying one. A modification overwrites the original event. Google specifically warns against modifying page_view, because it would stop collecting pageviews for other pages. Review the list of modified events and confirm none replaces a core event.
Assumption example. A cleaner has 1,000 sessions and 25 enquiry-form submissions in a month. With page_view marked as a key event, GA4 would report about 3,000 key events if visitors average three pages. With a thank_you event, it reports 25. Only the second number can be divided into ad spend.
Which Referrals and Filters Need Cleaning?
Check 6: unwanted referrals. A third-party payment page or booking widget can send visitors back to the site, and GA4 may credit the sale to that domain instead of the original source. Open the web data stream, choose Configure tag settings, then List unwanted referrals. Add each domain. One stream accepts up to 50.
Google adds a catch. A person who first arrived through a domain before it was excluded can still be credited to it on a later direct return, because of the last-non-direct-click model. Expect old credit to fade, not vanish.
Check 7: internal traffic. Open Admin, then Data filters. GA4 offers filters for developer traffic, internal traffic by IP address and web hostname traffic. Staff visits, owner checks and agency work all distort a small site. A local business with 40 conversions a month feels five fake ones.
Filters are permanent. Google states that excluded data is never processed and never reaches Analytics or BigQuery, and filters do not change history. Check the IP range twice before activating one.
Check 8: hostname filter. The web hostname filter excludes data from domains that are not yours, which helps when a staging site or a copied page sends events to the same property. If nothing unfamiliar appears in the hostname report, skip it.
Are the Time Zone and Currency Correct?
Check 9: reporting time zone and currency. Both are chosen when the property is created. The time zone decides which day an event belongs to: a visit on Tuesday in a visitor's time zone can count as Monday in the business's. Pick the zone of the business.
Changing the time zone later affects data going forward only. Google warns of a flat spot or a spike around the change, and advises against changing it more than once a day. If a chart shows an odd step on one date, check the change history before blaming the campaign.
Is Google Ads Linked and Tagged Correctly?
Check 10: the link exists. In Admin, under Product links, open Google Ads links. The person linking needs Editor access in Analytics and admin access in Google Ads. One property supports up to 400 links, and a link to a manager account counts as one. After linking, Ads data appears in GA reports within 48 hours.
Google also states that customers who link accounts are correlated with a 23% increase in conversions and a 10% reduction in cost per conversion. That is Google's own statement about a correlation, not a promise for one account.
Check 11: auto-tagging. Auto-tagging is needed to import Ads campaign and cost data into Analytics reports and to bring key events into Ads. It is on by default for new Ads accounts. In GA4 it also overrides manual UTM tags on Ads clicks, with no switch to reverse that. If campaign data looks empty, confirm both the link and auto-tagging before testing anything else.
Check 12: what the link turned into key events. When a property is linked with ads personalization on, Google marks these events as key events: add_to_cart, add_to_wishlist, begin_checkout, session_start, view_item, view_item_list and view_search_results.
session_start fires on every session, so a service business can end up with key events that equal sessions. Open the key events list, remove the ones that are not real goals, and keep the thank_you event from check 4.
For online shops there is one more item. The purchase event is a recommended event, and sending it populates the Ecommerce purchases report. If the report is empty, the event is not firing.
What Prompt Can Run the Audit?
An AI model cannot see a GA4 property on its own. It can review settings that are pasted into the chat as text or screenshots. The prompt below turns the 12 checks into a report. Replace the bracketed items first.
You are a GA4 auditor for a local business: [business type, city].
I will paste screenshots or text from these GA4 screens:
Data retention, Events (key events), Data filters, Referral
exclusions, Property settings (time zone, currency), Google Ads links.
Check these 12 items and answer PASS, FAIL or UNKNOWN for each:
1 Event data retention is 14 months (standard property options: 2 or 14)
2 Reset user data on new activity is set on purpose
3 At least one real key event exists (booking, call, form)
4 page_view is NOT a key event; a thank-you event is
5 No core event is overwritten by a modified event
6 Payment or booking domains are in unwanted referrals (max 50)
7 Internal traffic filter exists for the team's IP range
8 No unknown hostnames in the hostname report
9 Time zone and currency match the business
10 Google Ads is linked to the property
11 Auto-tagging is on
12 Key events added by the Ads link (session_start, view_item and
others) are reviewed and removed if not real goals
For every FAIL give: the screen path, the exact change, and the
risk if left alone. If something is not visible in my paste,
write UNKNOWN. Do not guess. Finish with the 3 fixes to do first.
Run it per property. UNKNOWN answers are useful: they show which screen still needs a screenshot. Treat every FAIL as a suggestion and confirm it on the screen before changing anything. Filters in particular cannot be undone.
What Should the First Hour Look Like?
Start with the checks that cannot be repaired later. Retention comes first, because a wrong setting silently loses data every month. Check 7 comes next, with one rule: confirm the IP range, then activate. Everything else can be fixed at any time without losing history.
Once after setup, after every site redesign or new ad platform, and then every quarter. The retention and filter checks need only one pass unless the property changes. The key event and referral checks are the ones worth repeating.
Can GA4 recover data that already expired?
No. When data passes the retention period it is deleted automatically on a monthly basis, and a lowered setting deletes older data at the next monthly process. Raising retention helps only for data that still exists. This is why check 1 comes first.
Is it safe to apply an AI model's audit fixes directly?
Treat the output as a to-do list, not a command. Confirm each FAIL on the real screen. Data filters are permanent, and modified events overwrite the originals, so those two need a second look before saving.
Why do key events look far higher than real bookings?
Usually one of two causes. Either page_view or another high-volume event is marked as a key event, or linking Google Ads with ads personalization added events such as session_start as key events. Compare the key event list with the actual goals.
Does GA4 need a paid tool to be audited?
No. Every setting in this checklist is in the free Admin area, and the prompt needs only pasted text or screenshots. Paid tools become useful when several properties or ad accounts have to be reviewed on a schedule.
Local marketing strategist with 10+ years at global agencies — OMD, Dentsu, GroupM, and BBDO. Now helping small businesses get the same data-driven edge. Based in Europe, working with clients in the US, UK, Australia, and beyond.